Problem Solving (MA2201) Week 9 Timothy Murphy December 9, 2011 1. A domino covers 2 squares on a chess-board. If two opposite corner squares on the board are removed, show that it is not possible to cover the remaining 62 squares with 31 dominoes. Answer: The squares will be the same colour, say white, so there will be 32 black squares left but only 30 white. Each domino covers one black and one white square; hence there is no way of covering them all. 2. Find all polynomials f (x) such that f (x2 ) = f (x)2 . Answer: Suppose f (x) has just one term, say f (x) = cxn Then c2 = c =⇒ c = 1 (if f (x) 6= 0). Now suppose f (x) has more than one term. Let the first two terms be f (x) = an xn + am xm + · · · . Then f (x2 ) = an x2n + am x2m + · · · , while the first two terms of the square are f (x)2 = a2n x2n + 2an am xn+m + · · · . Hence 2m = m + n, which is impossible. Hence the only non-zero solutions are f (x) = xn . 3. Show that if A, B, C are the angles of a triangle then tan A + tan B + tan C = tan A tan B tan C. Answer: Since C = π − (A + B), tan C = − tan(A + B) tan A + tan B =− . 1 − tan A tan B Hence tan C − tan A tan B tan C = −(tan A + tan B), ie tan A + tan B + tan C = tan A tan B tan C. 4. Prove that n X k=1 2n−1 X (−1)k 1 = . n+k k+1 k=0 Answer: If n = 1 the equation reads 1 1 =1− , 2 2 while if n = 2 it reads 1 1 1 1 1 + =1− + − . 3 4 2 3 4 Let us try to proved the identity by induction. We have seen that it is true for n = 1. Suppose it is true for n − 1. On passing to n we add 1 1 1 1 1 + = + − + n 2n − 1 2n 2n 2n − 1 to the left-hand side, while 1 1 + 2n − 1 2n is added to the right-hand side. Since these are equal, it follows by induction that the identity holds for all n. 5. Does there exist a non-zero polynomial f (x, y) such that f ([x], [2x]) = 0 for all real x. (Recall that [x] is the largest integer ≤ x.) Answer: If g(x, y) = y − 2x then ( 0 if n ≤ x < n + 1/2, g([x], [2x]) = 1 if n + 1/2 ≤ x < n + 1. Thus if h(x, y) = 2y − x − 1 then ( −1 if n ≤ x < n + 1/2, g([x], [2x]) = 0 if n + 1/2 ≤ x < n + 1. Hence f (x, y) = g(x, y)h(x, y) has the required property. 6. Evaluate Z 0 1 log(x + 1) dx. x2 + 1 Answer: Integrating by parts, the integral I = S − J, where S = [log(x + 1) arctan(x)]10 π = log 2 · , Z 1 4 arctan x J= dx x+1 0 7. Given a point O and a line ` in the plane, what is the locus of a point P which moves so that the sum of its distances from O and ` is constant? Answer: Take the line through O parallel to `, and the line through O perpendicular to this, as coordinate axes. Then OP = p x2 + y 2 , On the other hand, the distance of O from ` is d + y, where d is the distance from O to `. Thus the locus is given by the equation p x2 + y 2 + d + y = c, ie x2 + y 2 = (c − y − d)2 . This reduces to x2 = 2(d − c)y + (c + d)2 = 4ay1 , where a = (d − c)/2 and y1 = y + (c + d)2 /4a. Thus the locus is a parabola, symmetric about the y-axis. 8. Show that if an > 0 and limn→∞ an = 0 then the equation ai + aj + ak = 1 holds only for a finite number of triples i, j, k. Answer: 9. In how many different ways can 2n points on the circumference of a circle. be joined in pairs by n cords which do not intersect within the circle? Answer: This is a nice exercise in generating polynomials. Let the number of ways of joining 2n points in this way be an . Suppose the points are P1 , P2 , . . . , P2n in that order around the circle. Let us choose one point, say P1 . It is easy to see that there are an even number of points between 2 points joined by a chord. So P1 can be joined to any of the points P2 , P4 , . . . , P2n . Suppose P1 is joined to P2r . There are 2(r − 1) points between P1 and P2r , and it is clear that these points are joined in accordance with the same rules. Thus these points can be joined in ar−1 ways. Similarly the 2(n − r − 1) points between P2r and P1 (going the same way round the circle) can be joined in an−r−1 ways. Thus there are ar−1 an−r−1 ways of joining the remaining points if P1 is joined to P2r . (This also holds if P1 is joined to an adjacent point, P2 or P2n , provided we set a0 = 1.) Adding the contributions from the different end-points to the chord from P1 , an = a0 an−1 + a1 an−2 + · · · + an−1 a0 , for n ≥ 1. Now let us introduce the generating function f (x) = a0 + a1 x + a2 x2 + · · · . The relation above translates into f (x) − 1 = xf (x)2 , ie xf (x)2 − f (x) + 1 = 0. Solving this quadratic equation for f (x), √ 1 ± 1 − 4x f (x) = . 2x Since f (x) is a power-series in x, we must take the negative sign: √ 1 − 1 − 4x f (x) = . 2x By the binomial theorem (1/2)(−1/2)(−3/2) (1/2)(−1/2) (4x)2 − (4x)3 − 2! 3! 1 1 · 3 = 1 − 2x − (2x)2 − (2x)3 − · · · . 2! 3! (1 − 4x)1/2 = 1 − (1/2)4x + Thus an = 1 · 3 · 5 · · · (2n − 1) n 2 . (n + 1)! 10. A hole of diameter 1 is drilled through the centre of a sphere of radius 1. What is the volume of the remaining material? Answer: 11. Solve the simultaneous equations x+y+z =2 x2 + y 2 + z 2 = 5 x3 + y 3 + z 3 = 8. Answer: Note that x, y, z are the roots of the equation t3 − at2 + bt − c, where a= X x = 2, b = X xy, c = xyz. Now X X X ( x)2 = x2 + 2 xy. Thus X xy = 1/2(4 − 5) = −1/2. Similarly X X X ( x)3 = x3 + 3 x2 y + 6xyz. But X X X x xy = x2 y + 3xyz. Hence X x2 y = X X x xy − 3xyz, and so X X X X 3 3 ( x) = x +3 x xy − 3xyz. Thus xyz = −1. The cubic is therefore t3 − 2t2 − 1/2t + 1 = 0, ie 2t3 − 4t2 − t + 2 = 0. By observation, this has solution t = 2. Dividing by t − 2, 2t3 − 4t2 − t + 2 = (t − 2)(2t2 − 1). Thus the roots are 2, ± √12 ; and the solution to the equations is 1 {x, y, z} = {2, ± √ }. 2 12. Show that for any positive integers m ≤ n the sum 1 1 1 + + ··· + , m m+1 n when expressed in its lowest terms, has odd numerator. Answer: 13. The function f (x) satisfies f (0) = 1, f 0 (0) = 0 and (1 + f (x))f ”(x) = 1 + x for all real x. Determine the maximum value of f 0 (1), and the maximum and minimum values of f 0 (−1). Answer: 14. A group G is a union of 3 proper subgroups if and only if there is a surjective homomorphism G → K where K is the Klein 4-group. Answer: 15. Find all solutions in integers of the equation x2 = y 3 + 1. Answer: Challenge Problem Let a1 = 1/2, an+1 = an − a2n . Find a real number c for which the sequence bn = nc an has a finite limit, and determine this limit. Answer: By induction 1 0 < an ≤ . 2 Since an is decreasing, it follows that an converges to a limit `. From the defining relation, ` = ` − `2 =⇒ ` = 0, ie an → 0. It is helpful to turn to an analogical problem from differential calculus, namely: Solve the differental equation dy = −y 2 . dx Here we are replacing n with x, the sequence an with the function y(x), and an+1 − an with dy/dx. This equation can be written − dy = dx, y2 with the solution 1 = x + c, y ie y= 1 . x+c bn = 1 , n+1 b1 = 1 = a1 . 2 So let us try so that Also a2 = 1/4, b2 = 1/3, a3 = 3/16, b3 = 1/4, a4 = 39/256, b4 = 1/5. By the Mean Value Theorem, bn+1 − bn = − 1 , (n + θ)2 where 1 < θ < 2. Thus 1 1 < b − n < , n n+1 (n + 1)2 n2 ie b2n+1 < bn − bn+1 < b2n . As we shall see, it follows from this by induction that an ≤ b n . This holds for n = 1. Suppose it holds for 1 ≤ r ≤ n. Then ar − ar+1 = a2r ≤ b2r < br−1 − br . Adding these inequalities for r = 1, 2, . . . , n, a1 − an + 1 ≤ b1 − bn+1 , and so an+1 ≤ bn+1 . Similarly